Building Enhanced Neural Network Models to Predict Energy Storage Density of Composite Materials for Low-Temperature Thermochemical Energy Storage
Journal of Energy Resources Technology, Part A: Sustainable and Renewable Energy · 01.11.2025 · 1 (6)
DOI: 10.1115/1.4069225 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Energy Resources Technology, Part A: Sustainable and Renewable Energy, online erschienen am 20.08.2025. Verfasst von Nemalikanti Anand und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 43 Literaturangaben.
Abstract (Originalsprache)
Composite materials play a critical role in thermochemical energy storage (TCES) systems due to their ability to enhance the performance, efficiency, and longevity of these systems. TCES systems rely on reversible chemical reactions to store and release heat, making them an essential technology for applications such as renewable energy integration, industrial waste heat recovery, and building heating or cooling. Here are the main reasons composite materials are significant: (i) improved thermal stability, (ii) enhanced heat and mass transfer, (iii) reduction in reactant leakage, (iv) tailored reaction kinetics, (v) higher energy density and recyclability, and (vi) enhanced durability against environmental conditions. Current predictive models for energy storage density (ESD) of composite materials rely heavily on experimental methods, which are time-consuming and resource-intensive. This study used an enhanced artificial neural network (ANN), a convolutional neural network (CNN), eXtreme gradient boosting (XGBoost), and random forest (RF) models to improve prediction accuracy for ESD using a dataset derived from 54 experimentally proven samples. The dataset, comprising various composite sorbents, was processed using techniques such as one-hot encoding and a standard scaler for feature scaling. The ANN model yielded an R2 value of 0.98, with a mean squared error (MSE) of 123.51 and a mean absolute error (MAE) of 8.12. The CNN model yielded an R2 score of 0.97, an MSE of 216.7, and an MAE of 12. The XGBoost model yielded an R2 score of 0.925, an MSE of 738.02, and an MAE of 18.95. Similarly, the RF model yielded an R2 score of 0.892, an MSE of 1052.57, and an MAE of 22.69. These results highlight the effectiveness of the proposed models in accurately predicting ESD, providing a robust tool for optimizing TCES applications.
Zitieren
APA
Anand, N., Uppalapati, S., Padamurthy, A., Kumar, P. K., Kumar, G. N., & Vemanaboina, H. (2025). Building Enhanced Neural Network Models to Predict Energy Storage Density of Composite Materials for Low-Temperature Thermochemical Energy Storage. Journal of Energy Resources Technology, Part A: Sustainable and Renewable Energy, 1(6). https://doi.org/10.1115/1.4069225
Vancouver
Anand N, Uppalapati S, Padamurthy A, Kumar PK, Kumar GN, Vemanaboina H. Building Enhanced Neural Network Models to Predict Energy Storage Density of Composite Materials for Low-Temperature Thermochemical Energy Storage. Journal of Energy Resources Technology, Part A: Sustainable and Renewable Energy. 2025;1(6). doi:10.1115/1.4069225
BibTeX
@article{Anand2025,
author = {Anand, Nemalikanti and Uppalapati, Sudhakar and Padamurthy, Ankammarao and Kumar, Panchagnula Kishore and Kumar, G. Naveen and Vemanaboina, Harinadh},
title = {Building Enhanced Neural Network Models to Predict Energy Storage Density of Composite Materials for Low-Temperature Thermochemical Energy Storage},
journal = {Journal of Energy Resources Technology, Part A: Sustainable and Renewable Energy},
year = {2025},
volume = {1},
number = {6},
doi = {10.1115/1.4069225}
}
EUVIMED